**4. Discussion**

In this prospective study, we focused on suspected necrotizing enterocolitis NEC-1 preterm infants. NEC-1 phenotype has been poorly clinically investigated, with no data available on gu<sup>t</sup> microbiota, microbiome and fecal metabolome. By contrast, NEC-2 and NEC-3, more severe and established phenotypes, have been more characterized. As for clinical parameters, the increased cordon lactate levels we found in NEC-1 has been recently positively correlated to the development of enteropathy [23]. Hence, the hypothesis of hypoxic lesions in utero or during birth may not be excluded, and could even be predictive of neonatal morbidity. Importantly, the observed reduced enteral nutrition volume in NEC-1 is not a protective factor during the development of severe NEC (NEC-2 and NEC-3), but rather, it may lengthen hospitalization and infections risk [24]. Moreover, a recent multicentric study showed that a slow rate enteral feeding is associated with an increased risk of developing NEC-2 and NEC-3 [14].

NEC-1 children showed a high general variance for gu<sup>t</sup> microbiota and fecal metabolome, which is in line with a personalized microbiota and fecal metabolome profiles of preterm infant [25]. This is evident from PCA analyses in Figure 4 C and Figure 5 E, where the metabolome profiles of the healthy vs. NEC-1 children appear to diverge, but are still presenting some overlap, due to high intragroup variance. Both this evidence and the delayed intestinal colonization of preterm infants [26,27] may explain the lack of NEC-1-specific microbial group in the first 10-days of life. The analysis by periods of ten days of life revealed a divergence for both gu<sup>t</sup> microbiota and microbiome in NEC-1 by the third 10-days of life. In particular, the higher abundance of *Staphylococcus* in NEC-1 is in accordance with the early colonization by *Staphylococcus* bacteria of the intestine of preterm infants [28]. This datum suggests the third 10-days as an optimal time window to be targeted by antibiotics directed against bacterial species higher in NEC-1, such as *Staphylococcus*. However, in our study NEC-1 children who underwent glycopeptide and aminoglycoside therapy were more numerous than healthy children. Therefore, this evidence suggests that NEC-1 may be associated with glycopeptide and/or aminoglycoside-resistance, since NEC-1 gu<sup>t</sup> microbiota was characterized by an increase, and not a decrease, of *Staphylococcus.* Since aminoglycosides are active antibiotics against enterobacteria, their administration could delay intestinal colonization by Proteobacteria, and thus promote the implantation of resistant genera, such as *Staphylococcus* and *Streptococcus*. Based on this evidence, our data sugges<sup>t</sup> not to prolong antibiotic therapy beyond the first week of life in preterm infants, if inflammatory parameters have

been normalized. Furthermore, NEC-1 gu<sup>t</sup> microbiota profile was associated with ulcerative colitis and host genetic variation in the ANP32E gene, encoding a protein implicated in cortico-resistance during ulcerative colitis [22]. NEC-1 children showed increased exposition to antenatal corticosteroids compared to healthy children, even though a study has not identified antenatal corticosteroids as a NEC-inducing factor [29]. Despite Anp32e-deficient mice display no sign of disease [30], it has not to be excluded the role of Anp32e in a model of gu<sup>t</sup> inflammation mimicking ulcerative colitis. Hence, further studies are warranted on genetic factors of NEC. In terms of microbial functions, the intragroup analysis showed in the third 10-days of life a higher restriction enzyme activity in the NEC-1 gu<sup>t</sup> microbiome. This bacterial activity, directed against bacteriophages and enriched in the newborn intestine [31], suggests an increased virus activity and, hence, a virome dysbiosis, beyond a microbiota dysbiosis, during NEC-1 evolution. Notably, with regard to the time-point > 30 days, microbiota and microbiome changes are still present and, most importantly, di fferent from those observed at earlier time-points. This suggests that NEC-1 onset may have influenced the evolution of gu<sup>t</sup> microbiota. All these microbial data are associated with our observation about a change in fecal amino-acids, such as leucine and serine, confirming the association between gu<sup>t</sup> microbiota dysbiosis and a change in amino-acids metabolism [32].

Finally, several factors may a ffect the gu<sup>t</sup> microbiota of preterm infants, e.g., hospital regimens, even with regard to the putative use of probiotics to blunt NEC. In a very recent publication, Kurath-Koller et al. showed the e fficacy of some probiotics in improving the gu<sup>t</sup> microbiome of very low birth weight infants during the first two weeks of age, in a triple-center cohort study [33]. In our mono-centric study, the very low birth weight had an impact on the fecal metabolome of both healthy and NEC-1 children, but not on their gu<sup>t</sup> microbiota nor on their microbiome. This evidence underlines the importance of the hospital environment for the evolution of both an eubiotic and a dysbiotic, e.g., during NEC-1, gu<sup>t</sup> microbiota.
